MPEC problem formulations and solution strategies with chemical engineering applications

With the development and widespread use of large-scale nonlinear programming (NLP) tools for process optimization, there has been an associated application of NLP formulations with complementarity constraints in order to represent discrete decisions. Also known as mathematical programs with equilibrium constraints (MPECs), these formulations can be used to model certain classes of discrete events and can be more efficient than a mixed integer formulation. However, MPEC formulations and solution strategies are not yet fully developed in process engineering. In this study, we discuss MPEC properties, including concepts of stationarity and linear independence that are essential for well-defined NLP formulations. Nonlinear programming based solution strategies for MPECs are then reviewed and examples of complementarity drawn from chemical engineering applications are presented to illustrate the effectiveness of these formulations.     

Simultaneous mixed-integer dynamic optimization for integrated design and control

We consider strategies for integrated design and control through the robust and efficient solution of a mixed-integer dynamic optimization (MIDO) problem. The algorithm is based on the transformation of the MIDO problem into a mixed-integer nonlinear programming (MINLP) program. In this approach, both the manipulated and controlled variables are discretized using a simultaneous dynamic optimization approach. We also develop three MINLP formulations based on a nonconvex formulation, the conventional Big-M formulation and generalized disjunctive programming (GDP). In addition, we compare the outer approximation and NLP branch and bound algorithms on these formulations. This problem is applied to a system of two series connected continuous stirred tank reactors where a first-order reaction takes place. Our results demonstrate that the simultaneous MIDO approach is able to efficiently address the solution of the integrated design and control problem in a systematic way.     

Barrier NLP methods with structured regularization for optimization of degenerate optimization problems

Barrier nonlinear programming (NLP) solvers exploit sparse Newton-based algorithms and are characterized by fast performance and global convergence properties. This makes them especially suitable for very large process optimization problems. On the other hand, they are frequently challenged by degenerate and indefinite problems, which lead to ill-conditioned Karush–Kuhn–Tucker (KKT) systems. Such problems arise when process optimization models contain linearly dependent constraints, or the reduced Hessian is not positive definite at the solution. This can lead to poor solver performance and may preclude finding successful NLP solutions. Moreover, such optimization models occur in blending problems and NLP subproblems generated by MINLP or global optimization strategies. To deal with these difficulties we present a structured regularization strategy for barrier methods that identifies and excludes dependent constraints in the KKT system while leaving independent constraints unchanged. As a result, more accurate Newton directions can be obtained and much faster convergence can be expected for the KKT system over the conventional regularization approach. Numerical experiments with examples derived from the CUTE and COPS test sets as well as two nonlinear blending problems demonstrate the effectiveness of the proposed method and significantly better performance of the NLP solver.     

Interior-point decomposition approaches for parallel solution of large-scale nonlinear parameter estimation problems

Multi-scenario optimization is a convenient way to formulate and solve multi-set parameter estimation problems that arise from errors-in-variables-measured (EVM) formulations. These large-scale problems lead to nonlinear programs (NLPs) with specialized structure that can be exploited by the NLP solver in order to obtained more efficient solutions. Here we adapt the IPOPT barrier nonlinear programming algorithm to provide efficient parallel solution of multi-scenario problems. The recently developed object oriented framework, IPOPT 3.2, has been specifically designed to allow specialized linear algebra in order to exploit problem specific structure. This study discusses high-level design principles of IPOPT 3.2 and develops a parallel Schur complement decomposition approach for large-scale multi-scenario optimization problems. A large-scale case study example for the identification of an industrial low-density polyethylene (LDPE) reactor model is presented. The effectiveness of the approach is demonstrated through the solution of parameter estimation problems with over 4100 ordinary differential equations, 16,000 algebraic equations and 2100 degrees of freedom in a distributed cluster.     

一种求解化工动态优化问题的迭代自适应粒子群方法

智能优化方法因其简单、易实现且具有良好的全局搜索能力,在动态优化中的应用越来越广泛,但传统的智能方法收敛速度相对较慢。提出了一种迭代自适应粒子群优化方法（IAPSO）来求解一般的化工动态优化问题。首先通过控制变量参数化将原动态优化问题转化为非线性规划问题,再利用所提出的迭代自适应粒子群优化方法进行求解。相比传统的粒子群优化方法,该种迭代自适应粒子群优化方法具有收敛速度更快的优点,主要原因是：该算法根据粒子种群分布特性自适应调整参数;该算法通过缩减搜索空间并迭代使用粒子群算法搜索最优解。将提出的迭代自适应粒子群方法应用到多个经典动态优化问题中,测试结果表明,该方法简单、有效,精度高,且收敛速度比传统粒子群算法有显著提升。     

非线性动态优化问题的改进控制矢量迭代法(Ⅰ)无路径约束问题(英文)

This study proposes an efficient indirect approach for general nonlinear dynamic optimization problems without path constraints. The approach incorporates the virtues both from indirect and direct methods: it solves the optimality conditions like the traditional indirect methods do, but uses a discretization technique inspired from direct methods. Compared with other indirect approaches, the proposed approach has two main advantages: (1) the discretized optimization problem only employs unconstrained nonlinear programming (NLP) algorithms such as BFGS (Broyden-Fletcher-Goldfarb-Shanno), rather than constrained NLP algorithms, therefore the computational efficiency is increased; (2) the relationship between the number of the discretized time intervals and the integration error of the four-step Adams predictor-corrector algorithm is established, thus the minimal number of time intervals that under desired integration tolerance can be estimated. The classic batch reactor problem is tested and compared in detail with literature reports, and the results reveal the effectiveness of the proposed approach. Dealing with path constraints requires extra techniques, and will be studied in the second paper.     

A bilevel NLP sensitivity‐based decomposition for dynamic optimization with moving finite elements

Optimal control has guided numerous applications in chemical engineering, and exact determination of optimal profiles is essential for operation of separation and reactive processes, and operating strategies and recipe generation for batch processes. Here, a simultaneous collocation formulation based on moving finite elements is developed for the solution of a class of optimal control problems. Novel features of the algorithm include the direct location of breakpoints for control profiles and a termination criterion based on a constant Hamiltonian profile. The algorithm is stabilized and performance is significantly improved by decomposing the overall nonlinear programming (NLP) formulation into an inner problem, which solves a fixed element simultaneous collocation problem, and an outer problem, which adjusts the finite elements based on several error criteria. This bilevel formulation is aided by a NLP solver (the interior point optimizer) for both problems as well as an NLP sensitivity component, which provides derivative information from the inner problem to the outer problem. This approach is demonstrated on 11 dynamic optimization problems drawn from the optimal control and chemical engineering literature. © 2014 American Institute of Chemical Engineers AIChE J, 60: 966–979, 2014     

Nonlinear Optimization Strategies for Process Separations and Process Intensification

Advanced nonlinear programming (NLP) strategies based on equation-oriented (EO) process models are leading to significant improvements in computer-aided process engineering. The EO paradigm allows the development of large, integrated optimization platforms that expand the scope of continuous optimization tasks in process engineering. In particular, these platforms deploy significantly faster NLP strategies than in commercial simulation tools. Moreover, they exploit exact derivatives and system structure in order to consider much larger and more challenging systems. Finally, they allow the incorporation of much more general models, such as multi-level optimization and complementarity constraints. For process optimization this allows the treatment of extended models for complex phase equilibrium and process separations. These advances facilitate the optimization of novel integrated systems that arise in process intensification. Several separation case studies are presented that illustrate these optimization concepts and demonstrate their effectiveness for hybrid membrane/distillation separations and reactive distillation systems that typify novel systems in process intensification.     

Integrated Optimization Strategies for Dynamic Process Operations

Process systems engineering faces increasing demands and opportunities for better process modeling and optimization strategies, particularly in the area of dynamic operations. Modern optimization strategies for dynamic optimization trace their inception to the groundbreaking work Pontryagin and his coworkers, starting 60 years ago. Since then the application of large-scale non-linear programming strategies has extended their discoveries to deal with challenging real-world process optimization problems. This study discusses the evolution of dynamic optimization strategies and how they have impacted the optimal design and operation of chemical processes. We demonstrate the effectiveness of dynamic optimization on three case studies for real-world reactive processes. In the first case, we consider the optimal design of runaway reactors, where simulation models may lead to unbounded profiles for many choices of design and operating conditions. As a result, optimization based on repeated simulations typically fails, and a simultaneous, equationbased approach must be applied. Next we consider optimal operating policies for grade transitions in polymer processes. Modeled as an optimal control problem, we demonstrate how product specifications lead to multistage formulations that greatly improve process performance and reduce waste. Third, we consider an optimization strategy for the integration of scheduling and dynamic process operation for general continuous/batch processes. The method introduces a discrete time formulation for simultaneous optimization of scheduling and operating decisions. For all of these cases we provide a summary of directions and challenges for future integration of these tasks and extensions in optimization formulations and strategies.     

Advances in simultaneous strategies for dynamic process optimization

Following on the popularity of dynamic simulation for process systems, dynamic optimization has been identified as an important task for key process applications. In this study, we present an improved algorithm for simultaneous strategies for dynamic optimization. This approach addresses two important issues for dynamic optimization. First, an improved nonlinear programming strategy is developed based on interior point methods. This approach incorporates a novel filter-based line search method as well as preconditioned conjugate gradient method for computing search directions for control variables. This leads to a significant gain in algorithmic performance. On a dynamic optimization case study, we show that nonlinear programs (NLPs) with over 800,000 variables can be solved in less than 67 CPU minutes. Second, we address the problem of moving finite elements through an extension of the interior point strategy. With this strategy we develop a reliable and efficient algorithm to adjust elements to track optimal control profile breakpoints and to ensure accurate state and control profiles. This is demonstrated on a dynamic optimization for two distillation columns. Finally, these algorithmic improvements allow us to consider a broader set of problem formulations that require dynamic optimization methods. These topics and future trends are outlined in the last section.     

Control Vector Parameterization‐Based Adaptive Invasive Weed Optimization for Dynamic Processes

A novel optimal approach named invasive weed optimization‐control vector parameterization (IWO‐CVP) for chemical dynamic optimization problems is proposed where CVP is used to transform the problem into a nonlinear programming (NLP) problem and an IWO algorithm is then applied to tackle the NLP problem. To improve efficiency, a new adaptive dispersion IWO‐based approach (ADIWO‐CVP) is further suggested to maintain the exploration ability of the algorithm throughout the entire searching procedure. Several classic chemical dynamic optimization problems are tested and detailed comparisons are carried out among ADIWO‐CVP, IWO‐CVP, and other methods. The research results demonstrate that ADIWO‐CVP not only is efficient, but also outperforms IWO‐CVP in terms of both accuracy and convergence speed.     

An augmented Lagrangian interior-point approach for large-scale NLP problems on graphics processing units

The demand for fast solution of nonlinear optimization problems, coupled with the emergence of new concurrent computing architectures, drives the need for parallel algorithms to solve challenging nonlinear programming (NLP) problems. In this paper, we propose an augmented Lagrangian interior-point approach for general NLP problems that solves in parallel on a Graphics processing unit (GPU). The algorithm is iterative at three levels. The first level replaces the original problem by a sequence of bound-constrained optimization problems using an augmented Lagrangian method. Each of these bound-constrained problems is solved using a nonlinear interior-point method. Inside the interior-point method, the barrier sub-problems are solved using a variation of Newton's method, where the linear system is solved using a preconditioned conjugate gradient (PCG) method, which is implemented efficiently on a GPU in parallel. This algorithm shows an order of magnitude speedup on several test problems from the COPS test set.     

Global optimization of stiff dynamical systems

We present a deterministic global optimization method for nonlinear programming formulations constrained by stiff systems of ordinary differential equation (ODE) initial value problems (IVPs). The examples arise from dynamic optimization problems exhibiting both fast and slow transient phenomena commonly encountered in model-based systems engineering applications. The proposed approach utilizes unconditionally stable implicit integration methods to reformulate the ODE-constrained problem into a nonconvex nonlinear program (NLP) with implicit functions embedded. This problem is then solved to global optimality in finite time using a spatial branch-and-bound framework utilizing convex/concave relaxations of implicit functions constructed by a method which fully exploits problem sparsity. The algorithms were implemented in the Julia programming language within the EAGO.jl package and demonstrated on five illustrative examples with varying complexity relevant in process systems engineering. The developed methods enable the guaranteed global solution of dynamic optimization problems with stiff ODE–IVPs embedded.     

Inferential Control of Reactive Destillation Columns – An Algorithmic Approach

Decentralized control system design comprises the selection of a suitable control structure and controller parameters. Here, mixed integer optimization is used to determine the optimal control structure and the optimal controller parameters simultaneously. The process dynamics is included explicitly into the constraints using a rigorous nonlinear dynamic process model. Depending on the objective function, which is used for the evaluation of competing control systems, two different formulations are proposed which lead to mixed‐integer dynamic optimization (MIDO) problems. A MIDO solution strategy based on the sequential approach is adopted in the present paper. Here, the MIDO problem is decomposed into a series of nonlinear programming (NLP) subproblems (dynamic optimization) where the binary variables are fixed, and mixed‐integer linear programming (MILP) master problems which determine a new binary configuration for the next NLP subproblem. The proposed methodology is applied to inferential control of reactive distillation columns as a challenging benchmark problem for chemical process control.     

A hybrid optimization strategy for simultaneous synthesis of heat exchanger network

The heat exchanger network synthesis problem often leads to large-scale non-convex mixed integer nonlinear programming formulations that contain many discrete and continuous variables, as well as nonlinear objective function or nonlinear constraints. In this paper, a novel method consisting of genetic algorithm and particle swarm optimization algorithm is proposed for simultaneous synthesis problem of heat exchanger networks. The simultaneous synthesis problem is solved in the following two levels: in the upper level, the network structures are generated randomly and reproduced using genetic algorithm; and in the lower level, heat load of units and stream-split heat flows are optimized through particle swarm optimization algorithm. The proposed approach is tested on four benchmark problems, and the obtained solutions are compared with those published in previous literature. The results of this study prove that the presented method is effective in obtaining the approximate optimal network with minimum total annual cost as performance index.     

Improvement of state profile accuracy in nonlinear dynamic optimization with the quasi‐sequential approach

Quasi‐sequential methods are efficient and flexible strategies for the solution of dynamic optimization problems. At the heart of these strategies lies the time discretization and approximation of dynamic systems for nonlinear optimization problems. To address this question, we employ a time derivative analysis within the quasi‐sequential approach and derive a finite element placement strategy. In addition, methods for direct error prediction are applied to this approach and extended with a proposed time derivative analysis. According to the information for current time derivatives, subintervals are introduced that improve accuracy of state profiles. Since this is only done in the simulation layer, the nonlinear programing solver need not be restarted. An efficient gradient computation is also derived for these subintervals; the resulting enhanced accuracy accelerates convergence performance and increases the robustness of the solution to initialization. A beer fermentation process case study is presented to demonstrate the effectiveness of the proposed approach. © 2011 American Institute of Chemical Engineers AIChE J, 2011     

New developments and capabilities in prosyn—An automated topology and parameter process synthesizer

This paper describes an improved, equation oriented user friendly version of the modular computer package PROSYN—a mixed-integer nonlinear programming (MINLP) process synthesizer. A number of strategies are described and implemented for the outer approximation and equality relaxation algorithm (OA/ER) in order to reduce the undesirable effects of nonstructured nonconvexities in the master problem. These strategies include use of valid outer-approximations for splitters and mixers and various ways of handling the linearizations. An NLP initializer, model generator and a comprehensive library of models for basic process units and interconnection nodes, and a comprehensive library of basic physical properties have been developed and built in the PROSYN package. In this way PROSYN can be used to solve problems at different levels of complexity: from the simple NLP flowsheeting up to the MINLP synthesis through the modeling and decomposition (M/D) strategy and simultaneous heat integration including HEN costs. Applications with PROSYN are demonstrated with several example problems.     

基于信赖域算法的精馏塔优化

从结构优化角度建立精馏塔优化的混合整数非线性规划(MINLP)模型,为了消除整数变量,引入绕流效率将MINLP问题转化为非线性规划(NLP)问题。针对得到的NLP问题提出一种优化方法,在该方法中采用结构优化中常用的信赖域优化算法进行求解,并应用虚拟瞬态连续性方程辅助优化中的稳态模拟。采用提出的优化方法对3个精馏系统进行设计优化,以不同初始值开始,均可得到令人满意的优化结果,表明所提优化方法具有良好的稳健性,对于较复杂的部分热耦合精馏过程仍然可以有效优化求解;信赖域算法在精馏塔优化中也表现出良好的收敛性。     

A trust region filter method for glass box/black box optimization

Modern nonlinear programming solvers can be utilized to solve very large scale problems in chemical engineering. However, these methods require fully open models with accurate derivatives. In this article, we address the hybrid glass box/black box optimization problem, in which part of a system is modeled with open, equation based models and part is black box. When equation based reduced models are used in place of the black box, NLP solvers may be applied directly but an accurate solution is not guaranteed. In this work, a trust region filter algorithm for glass box/black box optimization is presented. By combining concepts from trust region filter methods and derivative free optimization, the method guarantees convergence to first‐order critical points of the original glass box/black box problem. The algorithm is demonstrated on three comprehensive examples in chemical process optimization. © 2016 American Institute of Chemical Engineers AIChE J, 62: 3124–3136, 2016